[jackal1-66]

Multiple things will be changed by this PR in the usage of the event pools:

• The pools are now chained so that when all the events are used in a file, another file will be pulled → no more repeated files when randomisation is enabled
• The development is a first step to use event pools on hyperloop
• Both an example and the test macro were implemented
  Fixes implemented along the way:
• RoundRobin not considered when using evtpools → now it is forwarded and the events are reused in case all the files have been opened and completely used
• The driver seed was being thrown away → with rngseed = 0 (the default) the generator called gRandom->SetSeed(0), discarding the driver seed. Now with 0 the seed is taken directly from o2-sim

[github-actions]

This PR did not have any update in the last 30 days. Is it still needed? Unless further action in will be closed in 5 days.

[jackal1-66]

It is needed, but to fix a few parts and solve the conflicts

[jackal1-66]

The code-review by Sandro has just been implemented. This covers:

• GeneratorFromO2Kine: reset the original event header when a file is
  closed, so no stale header leaks across a file boundary
• GeneratorFromO2Kine: close the current file only once the next one is
  known to be usable (a failed open no longer drops a good file)
• Expected number of events per generator (Generator::setExpectedNEvents,
  falling back to the job-wide count): inside a hybrid each sub-generator
  is told its share, computed from the fractions; the event pool passes
  it on to its internal reader. Used in the "not enough events" warning
  and in the "ran out of events" message (also in GeneratorHepMC)
• GeneratorHybrid: no lookahead event is scheduled after the last event
  of the job, and with fixed fractions a sub-generator is never asked for
  more than its share (request budget). Avoids a rare fatal "ran out of
  events" for pools holding exactly the requested events
• GeneratorHybrid::Init: a sub-generator whose Init() fails stops the job
• Unusable input (no readable file for extkinO2, no pool file for
  evtpool) is fatal at initialisation again; evtpool used to segfault
• rundpl.sh: set -o pipefail; README: roundRobin=false default and the
  behaviour above documented

Additional improvements were also done while fixing the issues before

• GeneratorHybrid: a failure of a sub-generator in a worker thread is
  handed to the main thread and raised there only if the event is used
  (with more than one worker the job used to hang in some of the validation tests I ran locally)
• GeneratorHybrid: negative fractions are now rejected
• GeneratorHybrid: call ROOT::EnableThreadSafety() before starting the
  workers; o2-sim-dpl-eventgen with several workers and two pools
  crashed. o2-sim already has that. Since this development is useful also on HY,
  having this on the hybrid generator is important. To be discussed if it's actually needed though

More tests with various scenarios were implemented, handling a few exceptional cases:

• test_EventPoolChaining: install a throwing fatal handler, new cases
  for the header reset, failed open, expected events, unusable input
• new test_GeneratorHybridEventPool (+ EventPoolTestUtils.h): hybrid
  scenarios for the lookahead, shares, budget, Init failures, HepMC,
  failure forwarding, cocktail and parallel modes

[alibuild]

Error while checking build/O2/fullCI_slc9 for a36ec38 at 2026-10-10 10:37:

## sw/BUILD/o2checkcode-latest/log
--
========== List of errors found ==========
++ GRERR=0
++ grep -a -v clang-diagnostic-error error-log.txt
++ grep -a ' error:'
/sw/SOURCES/O2/slc9_x86-64-slc9_x86-64/0/Generators/test/test_GeneratorHybridEventPool.cxx:141:10: error: escaped string literal can be written as a raw string literal [modernize-raw-string-literal]
++ [[ 0 == 0 ]]
++ exit 1
--

Full log here.

[alibuild]

Error while checking build/O2/fullCI_slc9 for ea60b56 at 2026-10-10 12:28:

## sw/BUILD/o2checkcode-latest/log
--
========== List of errors found ==========
++ GRERR=0
++ grep -a -v clang-diagnostic-error error-log.txt
++ grep -a ' error:'
/sw/SOURCES/O2/slc9_x86-64-slc9_x86-64/0/Generators/test/test_GeneratorHybridEventPool.cxx:141:10: error: escaped string literal can be written as a raw string literal [modernize-raw-string-literal]
++ [[ 0 == 0 ]]
++ exit 1
--

Full log here.

[alibuild]

Error while checking build/O2/fullCI_slc9 for 1cadd4a at 2026-10-12 05:06:

## sw/BUILD/o2checkcode-latest/log
--
========== List of errors found ==========
++ GRERR=0
++ grep -a ' error:'
++ grep -a -v clang-diagnostic-error error-log.txt
/sw/SOURCES/O2/slc9_x86-64-slc9_x86-64/0/Generators/test/test_GeneratorHybridEventPool.cxx:141:10: error: escaped string literal can be written as a raw string literal [modernize-raw-string-literal]
++ [[ 0 == 0 ]]
++ exit 1
--

Full log here.

[github-actions]

This PR did not have any update in the last 30 days. Is it still needed? Unless further action in will be closed in 5 days.